Research Of Online Inspection System For Knitting Needle Profile Based On Digital Image Processing

Dummy needles of computerized knitting machine is a small complex in structure and high size accuracy spinning parts,which the detection of is difficult.The automatic precise detection can significantly improve efficiency and reduce the production cost during the production process of dummy needles.Revolve around how to realize rapid consistency detection of the dummy needles of computerized knitting machine,this paper summarizes the development of detecting technology at home and abroad and the present research status,synthetically compares the advantages and disadvantages of various automatic detection method and put forward a solution of online inspection system for the consistency of knitting needle based on digital image processing.The general process of solution is like the followingafter processing the dummy needle image collected by detection system,the system get the information of dummy needle contour feature so as to calculate rapidly and accurately the straightness which shows the deformation of knitting needle,finally judge whether tested needle parts are qualified products.Aim at the main problem of the online inspection system based on digital image processing,subject completed the following main work:(1)Position repeatability problem of needle parts.In order to solve the position repeatability problem of needle under the camera lens,subject design a set of specially used carrying needle mechanism for knitting needle seeding,which each station effectively match with CMOS camera external trigger device;(2)Image denoising algorithm adaptability.Through analysis noise source and characteristics of needle image which the build system gather.this paper puts forward an adaptive median filtering algorithm.Compared with other algorithm in PSNR,operation duration and subjective evaluation,it is concluded that the proposed algorithm has good pertinence and adaptability;(3)Target contour orientation of needle image.First,operation by Canny operator ascertain roughly the outline of the needle image target edge,and by using the quadratic polynomial interpolation method accurate positioning target image contour edge in the subpixel level:(4)The establishment of inspection index and the evaluation model of inspection system.According to analysis of the characteristics of the knitting needle consistency,subject puts forward straightness of the position from transfer needle hook in the mouth to the needle back 2-d projection as evaluation index of its consistency and shows the calculated method by using the least squares fitting line.For the evaluation model of the inspection system,a mathematical model is proposed based on weighted Mahalanobis distance discriminant analysis,which synthetically judge whether the tested needle parts are eligible or not by using the effective degree index of threshold value choice during binanyzation the needle image.the index referring to knitting needle position and straightness index.